Incorporated herein is a computer program listing printout appendix of source code used to generate calibration sets and calibration transfer functions of a test pattern to enable calibration of halftone output images according to the present invention. Copyright, 1990, Softwest Software Inc. A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the "Program Printout Appendix", as it appears in the Patent and Trademark Office Patent file or records, but otherwise reserves all copyright rights whatsoever.
This invention relates to a method and apparatus for calibrating halftone output images from an imagesetting device. In particular, the present method and apparatus comprises steps and devices involved in producing a plurality of calibration sets and in using those calibration sets to adjust the uncalibrated halftone input images to produce calibrated output images as a function of image resolution, exposure intensity and screen frequency. Further, the calibration sets can be generated to produce halftone output images on either film or paper and for either positive or negative representation of the halftone input image.
Typically, there are four fundamental steps involved in producing typeset-quality output from a computer front end publishing system such as, e.g., a PostScript imagesetter (or typesetter). First, an application programed (a drawing or page layout package) is used to find what is wanted or actually printed on the paper or film. Second, the "print" option from the program is used to start the printing process. Typically, the print option translates what is seen on the screen into a page description language that the imagesetter can recognize. The page description language is what is then sent to the imagesetter.
Many conventional imagesetter systems consist of raster image processors (RIP) and recorders. An exemplary RIP is one sold under the Model No. RIP-30 by Linotype Company of Hauppauge, N.Y. The RIP performs the third step in the process by converting the page description language sent to it from the front end into a raster or a grid of lines. Each line contains a series of pixels, which can be considered as a list of dots which turn on or off to form the image. In the fourth step, the RIP sends this raster to the recorder, which uses a modulated light source to expose the image onto either photographic film or paper. An exemplary recorder is one sold as Model No. Linotype-330, also by Linotype Company. The photographic medium, in a light-tight takeup cassette, is then removed from the recorder and photo developed by a conventional chemical process.
All imagesetters have a light source intensity adjustment to provide proper exposure of the film. Exposed areas turn black on both photographic film medium or paper medium after development. The amount or degree of blackness, called "density" is a function of how much the film is exposed. Strictly speaking, density is a measure of how much light is blocked by a piece of film. Conversely, if photographic paper is used, density is the measure of how much light is not reflected. Higher density values correspond to a "blacker" piece of film.
These films are typically used in the publishing industry to make contact printing plates. For reasons of quality and ease of production in printing plate manufacture, it is desirable that the maximum density measurement on a film be in excess of 4.00. The maximum density available is a function of the light source intensity, the photographic medium and the process and chemicals used for development.
Since conventional printing processes cannot reproduce continuous tone tints or images, gray scales (tints) and halftones are approximated in page description languages by overlying a halftone grid on top of the recorder's pixel grid. The halftone grid, often called a screen, is specified by the number of halftone cells per inch (or lines per inch) and the angle from the vertical by which the halftone grid is rotated. Each cell of the half tone grid encompasses a specific number of pixels of the recorder's pixel grid. A continuous tone image is simulated by assigning a gray value to each halftone cell contained within the image. Thus, imagesetters which use PostScript.RTM. are unable to produce shades of gray directly, generally approximate shades of gray by utilizing the aforementioned halftone technique. Halftone grids overlying the top of the pixel grids allows for a shade of gray to be approximated by turning a certain percentage of pixels on. The gray value of the halftone is obtained by averaging the gray continuous tone values contained within the dimensional boundaries of the halftone cell. Each halftone cell approximates a shade of gray by turning on, in dot shape, a certain percentage of the pixels contained within the cell. As the cell needs to be darker, more of the pixels are turned on. Theoretically, a 50% gray tint is made by turning on 50% of the pixels in a half-tone cell. These conventional techniques are well known in the art.
Associated with halftones is the problem commonly discovered by densitometers. A densitometer is a scientific instrument that measures density, or the ability of a piece of film to block light. The blacker the film, the more light it absorbs, and the higher the density reading on the densitometer. When measured with a densitometer, a halftone area that was supposed to be 50% gray will typically (for paper positive) be darker, because the higher exposure intensity needed for a higher maximum density (Dmax) tends to expand the pixel size, causing the overall halftone-dot to be larger than it actually is. Thus, halftone images, when uncalibrated, produce a gray value which is typically darker than that which the gray value should be. That is, when 50% of the pixels within a halftone cell are exposed, the resulting dark area covers more than 50% of the area contained by the halftone cell. At a dot area value of 0%, none of the halftone cell's pixels are exposed; at 100% dot value area, all of the pixels are exposed. Both 0% and 100% are always obtainable, therefore a non-linear relationship exists between the commanded dot area (or gray value) and the dot area actually achieved. A series of tint boxes in 5% intervals between 0% and 100% is commonly called a "gray scale".
The non-linear response of the gray scale is a function of the resolution of the recorder pixel grid, the light source intensity, the photographic media (film or paper) and the screen grid. These, however, are not the sole contributors to gray scale non-linearity. Contributing factors include the recorder light source, different film batches and photoprocessor chemicals.
Page description languages referred to hereinabove usually contain a mechanism to correct for non-linearities in gray scale and halftone output. An exemplary page description language is one sold under the trade name PostScript.RTM., by Adobe Systems Inc. of Mountain View, Calif. The mechanism Adobe included in PostScript.RTM. is called the "transfer function", and it provides the lowest level means to calibrate the imagesetter to correct for non-linear gray scale response. Postscript.RTM. transfer functions suffers an important disadvantage in not having the ability to compensate for all the numerous variables which typically cause the non-linear response, e.g., resolution variation in media, resolution, screen frequency exposure intensity, etc.
Not only are conventional imagesetters, and transfer functions therein, lacking in necessary complexity so as to compensate for numerous variations in, e.g., the resolution of the image to be calibrated, the exposure intensity of the to-be-developed image, the screen frequency, etc., but conventional devices also cannot produce precision transfer functions applicable to color image processing. Current imagesetters are black and white devices and "process color" reproductions are produced by overprinting the colored inks yellow, cyan, megenta and black. The colors of an image are "separated" into yellow, cyan, megenta and black components and are then outputted on black and white film in precise percentages. The films, called color separations, represent the amount of process color needed in each area of the page, and are used to make the printing plates. The imagesetter produces black and white, and the ink used with the final printing plate produces color. The RIP and imagesetter mechanism that produces tints and halftones also produces color separations. The key difference is that each color separation is rotated with respect to the others and its respective line per inch specification is slightly different. The rotation and change in lines per inch specification avoids the dots lining up in a moire pattern. While gray scale calibration is mostly insensitive to rotation, it is a function of the halftone lines per inch.
Gray scale calibration is well known in the industry. A series of pre-determined gray scale tint boxes, at pre-determined dot areas, are requested of the RIP and imagesetter, through the page description language, and each tint box on the resulting film is measured with a dot-area densitometer described hereinabove. The resulting measurement, called a "transfer curve" is then used to adjust the requested dot area by an amount necessary to achieve calibrated output.
However, as mentioned above, conventional transfer curves suffer many inherent disadvantages. Existing art divides the method in which the gray scale calibration adjustments are effected into two categories: application program-based and RIP-based. In application program-based gray scale calibration, an attempt is made to compensate for the non-linearities by sending an adjusted transfer curve immediately before the print job. This process is tedious and error prone because it requires an operator to construct a transfer curve under identical output conditions to the present, and to properly identify and select a specific curve to be sent with the print job. This process is duplicated for each application program, and for each imagesetter. If there are multiple front ends sending the imagesetter print jobs, then the files containing the curves must be distributed and managed. The number of transfer curves can become unruly. In an effort to reduce the number of transfer curves required, conventional methods require limiting the number of output conditions which can be specified. For instance, if film negatives at high resolutions of 150 lines per inch was the only output condition, then only one transfer curve per application, per imagesetter would be required. This is typically the case for closed reproduction systems, such as that mentioned in U.S. Pat. Nos. 4,075,662 and 4,500,919. The limitation on output conditions, however, is unrealistic and impractical with multi-function RIPs and imagesetters such those used in Model Nos. RIP-30 and Linotronic-330. A typical production environment using this equipment may require 24 transfer curves, for 5 applications, for 2 imagesetters. This would yield 240 separate transfer curves (24.times.5.times.2=240). If these curves were distributed among four front ends, then the operator would have to manage 960 separate files. In practice, the number of transfer curves used is a balance of output flexibility versus the number of curves to manage.
The second method by which conventional calibration adjustments are made, or RIP-based calibration, was formed as an attempt to reduce the number of curves and files required to support a minimum of output flexibility. Thus, RIP-based calibration is an approach used to calibrate the RIP and the imagesetter as a unit, so that all front ends and all application programs assume the device produces a linear gray scale. This process requires an operator to construct a transfer curve under identical output conditions to the present, and to properly identify and select that curve for use in the RIP and imagesetter. In the existing art, the operator loads the single curve into violative memory within the RIP. This is not repeated for each application programmable front end. In our example, this has the net effect of reducing a number of transfer curves needed from 240 to 48, and the number of files from 960 to 48. Although RIP-based calibration is effective in reducing the number of transfer curves, current RIP-base calibration techniques do not utilize many of the variable inputs which can effect the transfer curve. Furthermore, conventional RIP-based calibration requires manual selection and loading of the transfer curve appropriate to the current output conditions. Manual selection and loading for each transfer curve is not only inefficient but difficult to incorporate and thus, it is error-prone.
Accordingly, current calibration methodology requires that separate transfer curves be constructed to account for the plurality of variable inputs. Included with those inputs is either positive or negative output sense conditions. Images can be outputted in either a positive or negative format, one format or sense being the inversion of the other. Images, lines and type in a page description language are usually specified from the front end in a positive sense. The RIP and imagesetter can output any positive sense, or can switch to a negative independent of the page description. The negative sense is useful in many printing processes. Conventional methods for specifying a negative output sense are as follows:
1. Mathematically, through the transfer function and its interaction with the page description language. This method converts a gray value to negative by subtracting the gray value from 100 prior to halftone dot generation. Hence, a 0% gray value becomes 100%, 25% becomes 75%, etc.
2. Electronically, by inverting the pixel stream in the RIP prior to it leaving the RIP. A gray value is converted to a raster-based halftone dot, the halftone dot is placed in the RIP's raster memory, and after the complete page description is processed, the raster memory is transferred from the RIP in serial fashion to the recorder. Each bit in the pixel stream is inverted; that is, if the bit were on, it is turned off, if the bit were off, it is turned on. This produces a negative image, but because it has the effect of "turning the halftone dot inside out" the transfer curve required to produce the negative calibrated output is substantially different from that used with the mathematical negative.
3. Electronically, by inverting the pixel stream in the recorder immediately after it enters the recorder. This method is substantially similar to the second method above, and uses the same transfer as the second method in order to calibrate.
In practice, these methods are multiplicate in nature. That is, each of the three output sense settings shown above are set either positive or negative, and combine to produce an image. The negative of a negative is a positive. For example, the first method shown above places the output sense setting as positive, and the second and third method each set output sense setting as positive and the resultant image is positive. However, if method one places sense setting as negative, but method two places sense as negative and method three as positive, the resultant image is also positive. The transfer curve for these two cases, however, are different and this must be recognized by the operator for successful calibration. Thus, regardless of which method is used or whether or not the methods are multiplicated such that more than one method is used, an operator must set the methodology each time positive or negative output sense conditions are specified. Accordingly, conventional methods shown above which set sense output are often prone to operator error.
The net result of conventional art and all the complexities described hereinabove, is that halftone calibration for imagesetters with multiple output conditions such as, e.g., resolution, media, halftone frequency, screen intensity and sense/polarity output, is non-intuitive, labor-intensities, tedious and therefore error prone. Current methods require an inordinate amount of attention from the operator for success.